Ending of activation group
The IBM® i Integrated Language Environment® (ILE) program model uses activation groups as a way to encapsulate resources for an application program within a job.
When multiple threads exist in the jobs that possibly use the activation group, the system cannot end that activation group in a predictable and safe way. To solve this problem, the system changes the behavior of ending an activation group in a multithreaded job.
In IBM i
6.1 and earlier releases, if a secondary thread in a multithreaded
application performs an action that ends an activation group (for
example, C functions exit() or abort()), or fails to prevent an activation
group from ending, the system ends the job at that time. Beginning
with IBM i 7.1, the job ends
only if an activation group is ended in a secondary thread and the
activation group is active in more than one thread.
In IBM i 6.1 and earlier releases,
if the initial thread in a multithreaded application performs an action
that ends an activation group, or fails to prevent an activation group
from ending, the system does not end the job. Beginning with IBM i 7.1, if the initial thread
in a multithreaded application performs an action that ends an activation
group, or fails to prevent an activation group from ending and that
activation group is active in more than one thread, the system ends
the job.
A typical example occurs when a *NEW activation
group program is called from a secondary thread. When the program
ends, the *NEW activation group is destroyed. In IBM i 6.1 and earlier releases, the job ends.
If an HLL exit verb, like the C Runtime exit() function,
is called from a secondary thread in IBM i
6.1 and earlier releases, the job ends. Beginning with IBM i 7.1, the job ends only if the activation
group being ended is active in more than one thread.
See Activation groups and threads for more
information about this topic.